Emergent capabilities in large language models (LLMs) refer to the sudden and unpredictable development of new abilities as model size, computational power, and training data increase, leading to breakthroughs in tasks previously deemed impossible. This phenomenon presents both opportunities and challenges for AI researchers, who must balance current capabilities with the potential for future innovations by imagining new problems to solve. For instance, DeepL's Clarify tool showcases how emergent capabilities can enhance human-AI collaboration and personalize translations, adapting to individual user preferences while maintaining translation quality and accuracy. As LLMs continue to evolve, strategic investments in potential capabilities could revolutionize how language is translated across various cultures and languages, emphasizing the importance of planning for these possibilities rather than relying solely on existing methods. Stefan Mesken, Chief Scientist at DeepL, highlights the need to embrace the complexity and nuances of language by pushing the boundaries of current technologies to navigate cultural diversity without oversimplification, as emergent capabilities invite a new era of AI-driven communication.